Open distortion project

[mystran]

some opamps have non-linear inputs which may be uncorrected. for example the common OTA ca-3080 has huge non-linear response when driven at reasonable levels. you need to feed linearization current which most designs do not.

Yeah well, OTAs are kinda different from regular op-amps anyway, as you would normally use an OTA like 3080 in open-loop configuration. Sure you see regular OPAs used that way too, but most of them got enough gain that you're probably just railing it immediately (on purpose of course), no?

there are also opamp issues that can cause locking, inversion, reversal, oscillation and others. you'd need to ask a wizard familiar with opamp torture for specific tricks but i've seen them used intentionally here and there.

Urgh.

Well, at least such designs aren't very common, fortunately.

[camsr]

Let's put it this way: Feedback happens faster than sample rate, and in between samples there were many feedbacks that did something. But in digital, unless you EXPLICITLY define this behavior, the feedback is discrete time based, and not going to happen the same way. To get this behavior, requires more CPU time between samples, and depending on how large the circuit is, CPU will eventually cease to process the audio real-time.

I am sure someone could provide a simple, more intelligent example of this then I could

So you basically have to solve a delayless structure for non-linear feedback, I've seen quite a couple of examples here on kvr of delayless feedback filters but none of them solve for non-linear feedback

The way I see it is, to process the feedback in one sample frame, the process operation must be done some number of times within the frame. Eventually the feedback is killed for that frame because we must not process any more samples for many reasons, such as noise, lack of SNR, or lack of effect. The more CPU operations in the process, the more usage, as it multiplys the time the operation took by the number of iterations of it.

In reality, if the sample rate is increased enough to make this feedback feasible within the time frame of the effected frequency band, then delayless feedback should not be required, unless there is something I don't know about it. If comb filtering can be avoided, then sample delayed feedback should be a solution.

[Ichad.c]

Good point Camsr. Searching around here @ KVR I found a couple suggestions that I find quite interisting and is actually the kind of studd/examples I'm looking for, even though they are geared more towards filters: These are quoted from mystran's ideas(thanks!) on this thread:

http://www.kvraudio.com/forum/viewtopic ... &start=120

Even though they are geared more towards filters

Personally I've used something like (and I'm speaking from my head so might make mistakes):

lp += g*waveshape(in - lp);

lp = waveshape( lp + g*(in-lp) );

lp = waveshape( (1-g)*lp ) + g*in;

"Everyone of those sounds different. There are probably more. And most definitely andy has a valid point that if one is trying to model a specific circuit, then it is necessary to analyze where and how the non-linearities make their effects, instead of just dropping some shaping in some random place (and the most boring two of those random places being input and output)."

Mystans's statement above is exactly why I started this thread.

Regards
Andrew Ainslie

[Ichad.c]

By statement I mean:

...most boring two of those random places being input and output.

Emphasis on boring.

Regards
Andrew Ainslie

[aciddose]

yes considering the majority of non-linear effects in most filters i've seen are the result of non-linear V->I stages driving integrators (the effect of OTAs i described) the only way to get this result is to apply the same shaping at the same point. doesn't work at input/output because very little happens there. the only significant i/o related effect in most filters is the intentional clipping of the feedback to control oscillation.

this applies to simple circuits like the RDC where you have a resistor input and parallel capacitor and diode.

the biggest challenge is dealing with the computational complexity of some of the required non-linear functions and their resulting aliasing.

[mystran]

Oh and btw, keep in mind the above thread is like 4 years old.

[Ichad.c]

On the static waveshaper front. Does anybody know how to make a waveshaper that will retain the sharp edges of a triangle-wave and not turn it into something s-looking? Think that I've done it before - but I can't find the dang file anymore. Think I might have deleted it accidently